Mapping Contoured Terrain Using SLAM with a RadioControlled Helicopter Platform
Project Proposal
Cognitive Robotics, Spring 2005
Kaijen Hsiao
Henry de Plinval
Jason Miller
Introduction
In the context of a natural disaster, or when a military pilot has to eject in enemy
territory, Search and Rescue teams often have to find people in unknown or hazardous
areas. For safety reasons, Search and Rescue teams of the future will probably make use
of unmanned aerial vehicles. For such a rescue vehicle, the ability to localize itself, both
to avoid dangers (mountains/enemy bases) and to scan the entire area until survivors are
found, is essential. However, this area may be unknown (enemy territory), or not mapped
precisely (mountain summits). Moreover, global positioning systems (GPS) may not be
usable in the area, or they may be not accurate enough, as is often the case in areas with
dense foliage.
In such a situation, a helicopter capable of mapping its surroundings while
localizing itself on this map would be of special interest. In this project, we want to
investigate such a platform. We would like to implement a SLAM algorithm for a
helicopter moving in an area with uneven terrain and using 3-D rangefinder sensors. Our
goal is for this helicopter to be able to create a 2-D contour map of the ground
surrounding it, while localizing itself on that map.
Problem Statement
The goal of this project is to be able to do 2-D SLAM in a simulated, forested
outdoor environment where the ground is not flat. Our platform of choice is a small,
radio-controlled helicopter. In such a situation, traditional 2-D SLAM is problematic
because the horizontal plane of the laser rangefinder can hit contoured ground, causing
spurious landmarks to be placed on the map. It can also miss low-lying landmarks if the
helicopter is hovering too high. Thus, the objectives of this project are:
1) To simulate an appropriate forested outdoor environment and the
motion/perception of a small, radio-controlled helicopter
2) To use 3-D rangefinder data to create a leveled 2-D map, rejecting spurious
landmarks
3) To attach elevation data to points on the leveled 2-D map, creating a contour map
Previous Work
In terms of mapping of non-flat terrain from a helicopter, (Thrun, 2003) creates a
3-D map using 2-D rangefinder data. A small helicopter is equipped with rangefinders
whose measurements lie in a plane perpendicular to the direction of motion. Using scanalignment techniques, the noisy data is combined into a smooth 3-D picture of the world.
However, they are not using SLAM, and the helicopter cannot image the same location
twice with their algorithm.
(Montemerlo, 2003) creates a 3-D map of a non-flat underground mine from a
cart platform. It uses a forward-pointing vertical rangefinder (whose plane is parallel to
the direction of motion and the up-direction) to reject spurious ‘wall’ detections made by
a horizontal rangefinder pointed at non-level ground. The resulting data is used to create
a 2-D map with a normal 2-D SLAM algorithm. The 3-D map is then created using a
plane of rangefinder measurements perpendicular to the direction of motion, combined
using the robot’s estimate of its location on the 2-D map generated using SLAM and
smoothed using scan-alignment. This work is similar to what we are attempting to do, in
that it performs 2-D SLAM with disambiguation of spurious measurements due to
contoured terrain. However, we do not feel the need to create a full 3-D map, and the
leveled 2-D map they create uses only a single plane of vertical measurements. This is
sufficient under the assumption that the world is reasonably rectilinear, consisting only of
walls and ground. However, this is insufficient for outdoor environments.
Our project is essentially an extension of (Brenneke, 2003), which uses a
motorized cart equipped with a rotating 2-D laser rangefinder, just as described in our
proposed project, to map a contoured outdoor environment. The paper describes how to
use the 3-D cloud of points to disambiguate ground from landmarks in order to create a
leveled 2-D map. The techniques we will be using are largely similar to those used in
this paper. However, our extensions are twofold: first, because we are using a helicopter
platform, data can be taken from varying distances from the ground, and second, we hope
to add elevation data to the leveled 2-D map, which this paper does not do.
Outline of the Proposed Technical Approach
Each point in the problem statement will be addressed in turn.
1) To simulate an appropriate forested outdoor environment and the motion/perception
of a small, radio-controlled helicopter
The simulation will be done in Open Dynamics Engine, an open-source physics engine.
By applying the appropriate noisy forces to a simulated helicopter, we can simulate
motion of the helicopter using the physics engine. By detecting collisions between rays
protruding from the helicopter and the simulated world (and adding noise), we can create
a realistic picture of what a laser rangefinder might return. In order to eliminate having
to create autonomous navigation algorithms for the helicopter, we will assume that the
helicopter has a camera attached, allowing a human to drive it around.
2) To use 3-D rangefinder data to create a leveled 2-D map, rejecting spurious
landmarks
Our simulated helicopter will be equipped with a 2-D laser rangefinder pointed vertically
so that its measurement plane is parallel to both the up direction and the helicopter’s
direction of movement. To obtain 3-D data from this sensor, we will sweep it quickly
back and forth. Because the helicopter will continue to move during the sweep, the data
must be adjusted for the motion of the sensor. The 3-D data will be used to distinguish
ground and overhanging objects from legitimate landmarks, defined as objects sticking
directly out of the ground, such as tree trunks and rocks. Using these segmented
landmarks, a leveled 2-D map can be created using 2-D SLAM techniques. In particular,
we plan to use a particle-filter based SLAM algorithm, such as the FastSLAM algorithm
developed by Sebastian Thrun’s lab.
3) To attach elevation data to points on the leveled 2-D map, creating a contour map
To simplify the control of the helicopter, we will likely assume that it will try to hover a
fixed distance from the ground directly below it. Using the resulting inputs to the
helicopter and the 3-D data from the helicopter’s direction of travel, we can track an
estimate of the helicopter’s vertical movement. This would allow us to attach estimated
elevation data to each point on the leveled 2-D map.
Plans for Spirals 1 and 2
Spiral 1:
Create the simulation of the world and the robot, starting with the ground being flat.
Simulate the imperfect motion and 3-D perception of the helicopter, according to
appropriate motion/perception models.
Figure out how to adjust the data for movement of the robot.
Create a 2-D landmark detector.
Implement 2-D, particle-filter SLAM.
Spiral 2:
Make the simulated land contoured.
Extend the landmark detector to operate with 3-D data.
Track the helicopter’s vertical movement and attach elevation estimates to the map.
Planning
Division of labor
Kaijen: implement the simulated world, including obstacles, motion, and sensors; help
with algorithms to distinguish landmarks from ground and track elevation
Henry: implement the particle-filter SLAM algorithm; help with algorithms to distinguish
landmarks from ground and track elevation
Jason: create movement and sensor models; lead the implementation effort for the
algorithms to distinguish landmarks from ground track elevation
Time schedule:
April 10th-April 23rd (2 weeks):
Figure out how helicopter will move and motion/perception models
Create flat simulated world with helicopter and objects
Create basic landmark detector using 2-D subset of 3-D rangefinder data
Implement 2-D, particle-filter SLAM
Week of April 24th:
Create contoured terrain and 3-D rangefinder data
Implement ground/landmark separator using 3-D rangefinder data
Week of May 1st:
Create a tracker of the vertical movement of the helicopter using 3-D rangefinder data in
direction of motion
Attach elevation estimates to leveled 2-D map
Write presentation, paper
May 9th/11th: Presentation and report due-date
Bibliography
Brenneke, Christian, Oliver Wulf, Bernardo Wagner. “Using 3D Laser Range Data for
SLAM in Outdoor Environments,” IEEE/RSJ Conference on Intelligent Robots
and Systems, Oct. 2003.
Montemerlo, Michael, Dirk Hahnel, David Ferguson, Rudolph Triebel, Wolfram
Burgard, Scott Thayer, William Whittaker, and Sebastian Thrun. “A System for
Three-Dimensional Robotic Mapping of Underground Mines.” In IEEE
International Conference on Robotics and Automation (ICRA), 2003.
Thrun, Sebastian, Mark Diel, Dirk Hahnel. “Scan Alignment and 3-D Surface Modeling
with a Helicopter Platform,” The 4th International Conference on Field and
Service Robotics, 2003.